The "D20 vs. . . ." post got me thinking about what makes particular dice a good fit for one game or another.
d4: The standard d4 is an enigma for people who haven't been initiated into gaming through RPGs like D&D. The 25% odds on a particular result don't make for a very broad spread. It rests flat and stable on the tabletop, but there is also the caltrop factor for dice that end up on the floor, as stepping on one hurts like a bitch. Best with custom or color-coded faces. It lends itself to a sci-fi or mystic aesthetic.
d6: Everyone knows what a six-sided-die is. The only thing that sets players apart are the segment who view it as a mysterious and arbitrary random fetish versus the crowd who can't help but calculate every roll in 16.66% increments. These are the most cost effective option with the greatest variety of options available.
d8: This die seems to be getting some love lately with FFG. It's a weird shape though for the casual gamer and looks unbalanced and perhaps "strange" on the tabletop. The 12.5% increments are nice for throwing off mental calculations.
d10: The poor-man's percentile. These should be endemic to game designs, but they don't look good on the tabletop. The smaller faces and "0" proxy for the "10 value are a turn-off for casual gamers.
d12: I don't quite understand why this die hasn't seen more use. It's 2d6 rolled into one with generous faces imprinted with legible numbers. It rests evenly on the tabletop with handy 8% increments.
d20: It is hard to escape thoughts of D&D when envisioning this die. A fantastic die with 5% increments, the only real drawbacks are its relatively small faces and its imposing scope for casual gamers.
d30: This sucker is a monster, more like rolling a golf ball across the table, and I've only seen one game try to make use of it. The d100 is even less practical.
Take a look at these amazing non-conventional polyhedral dice (especially the preview video). The lite dice game they made for it really utilizes the dice designs.
Ugh. It's a fun shtick, but wildly impractical. I could see these working if you spun them like a top, but rolling something like that will never yield a good random spread. They would look nice packaged with an early '90s MB or TSR style game though. I picture in a promo ad with overly excited preteens sitting around a game board with the tagline, "Who will reach the wizard's tower first to seize his rod of rulership?"
They called you mad! MAD, I say! But you showed them all! The biggest obstacles with specialty dice like a d7 are just sourcing and cost. d6 cost pennies to include with plenty of choices for distributors and manufacturers, whereas d7 retail at $4 a pop from Chessex and Gamescience is charging $9 for an unattractive pair. I don't see them stocked by the manufacturer I used to order from. Whether it's those unique non-polyhedral dice above or specialty polyhedrons like the d7, it takes a company like Hasbro to make inclusion in a mass produced game cost effective.